Explainable Edge AI Framework for Real-Time Biomedical Waste Classification and Decision Support

Authors: Mr. A.Backiyaraj, S.Akshaya, W.Immanuel, K.Subbulakshmi, S.Venkat, B.Sourna Dharshini

Abstract: Biomedical waste management has become a critical challenge in modern healthcare due to the increasing volume of hazardous waste generated by hospitals, laboratories, and medical facilities. Traditional waste segregation methods are predominantly manual, making them prone to human errors, contamination risks, and inefficient disposal practices. Although cloud-based artificial intelligence (AI) systems have improved waste classification accuracy, their dependence on internet connectivity introduces latency, privacy concerns, and limited real-time applicability. This paper presents an Explainable Edge AI Framework for Real-Time Biomedical Waste Classification and Decision Support, which integrates lightweight deep learning models with edge computing to enable fast, accurate, and secure waste classification directly on embedded devices. The proposed framework incorporates Explainable Artificial Intelligence (XAI) techniques, including Grad-CAM and SHAP, to provide transparent visual explanations for each prediction, thereby enhancing user trust and supporting informed decision-making. In addition, an intelligent decision support module recommends appropriate segregation and disposal procedures according to biomedical waste management guidelines. The framework is designed to minimize inference time, reduce computational overhead, and maintain reliable offline operation in resource-constrained healthcare environments. Experimental evaluation demonstrates improved classification performance, enhanced interpretability, and efficient edge deployment, making the proposed system a practical and sustainable solution for smart healthcare waste management.